Sam Altman learned on Saturday evening that working with the U.S. government is currently a delicate undertaking. At around 7 p.m., the OpenAI CEO posted on X that he would be taking public questions, aiming to clarify his company's choice to accept a Pentagon contract that Anthropic had just abandoned. The inquiries largely centered on OpenAI's openness to involvement in mass surveillance and automated lethal operations—precisely the activities Anthropic had excluded during its own discussions with the Pentagon. Altman repeatedly deferred to the public sector, stating it was not his place to establish national policy. "I very deeply believe in the democratic process," he wrote in one reply, "and that our elected leaders have the power, and that we all have to uphold the constitution."

Just an hour later, he expressed surprise at the level of disagreement he encountered. "There is more open debate than I thought there would be," Altman noted, "about whether we should prefer a democratically elected government or unelected private companies to have more power. I guess this is something people disagree on."

This episode is revealing for both OpenAI and the broader technology sector. During the Q&A, Altman adopted a posture common in defense circles, where military and industry figures are expected to follow civilian authority. More telling, however, is that as OpenAI evolves from a highly successful consumer startup into a component of national security infrastructure, the company seems ill-prepared for its new duties. Altman's public forum occurred at a tense moment for the firm. The Pentagon had recently barred OpenAI competitor Anthropic for demanding contractual restrictions on surveillance and automated weapons. Shortly after, OpenAI revealed it had secured the very contract Anthropic relinquished.

Altman framed the agreement as a swift method to reduce tensions—and it was undoubtedly a profitable one. Yet he appeared caught off guard by the strong negative reaction from both users and staff. OpenAI has interacted with the U.S. government for years, but never under these circumstances. For example, when Altman testified before Congressional committees in 2023, he was largely employing a social media strategy. He spoke grandly about the company's transformative potential while recognizing risks and eagerly interacting with legislators—an ideal mix for attracting investors while forestalling regulation.

Less than three years later, that tactic is no longer sustainable. artificial intelligence is undeniably powerful, and the capital requirements are so substantial that a more serious partnership with the government is unavoidable. What is surprising is how unready both parties appear to be for this shift.

The most pressing immediate conflict involves Anthropic itself, following U.S. Defense Secretary Pete Hegseth's announcement on Friday of plans to label the lab a supply chain risk. This threat hangs over the entire discussion like a loaded weapon. As former Trump administration official Dean Ball noted over the weekend, such a designation would sever Anthropic's access to hardware and hosting partners, effectively crippling the company. It would be an unprecedented action against an American firm, and while it might eventually be overturned in court, the interim damage would be severe, sending ripples throughout the industry.

According to Ball's account, Anthropic was fulfilling an existing contract under terms agreed upon years earlier, only to have the administration demand changes to the agreement. This goes far beyond what would be acceptable between private entities and delivers an intimidating message to other contractors. "Even if Secretary Hegseth backs down and narrows his extremely broad threat against Anthropic, great damage has been done," Ball wrote. "Most corporations, political actors, and others will have to operate under the assumption that the logic of the tribe will now reign."

This poses a direct danger to Anthropic, but it also creates a significant challenge for OpenAI. The company is already facing strong internal pressure from employees to uphold certain ethical boundaries. Simultaneously, right-wing media will be watching closely for any indication that OpenAI is less than a firm political ally. At the center of it all is the Trump administration, working to make the situation as complex as possible.

One could argue that OpenAI never intended to become a defense contractor, but due to its enormous ambitions, it has been compelled to compete in the same arena as firms like Palantir and Anduril. Gaining traction during the Trump administration necessitates choosing sides. There are no neutral parties in this space, and gaining allies will inevitably mean creating opponents. It is still unclear what cost OpenAI will bear, whether in lost business or departing staff, but it is unlikely to escape without consequences.

It may appear odd that this pressure is intensifying at a time when more prominent tech investors hold influential roles in Washington than ever before, yet most seem perfectly content with tribal allegiances. Among venture capitalists aligned with Trump, Anthropic has long been viewed as excessively cozy with the Biden administration in ways that could harm the wider industry—a perception reinforced by Trump advisor David Sacks' response to the ongoing dispute. Now that the situation has reversed, few seem willing to defend the broader principle of free enterprise.

This is a challenging position for any organization. While politically aligned entities may gain short-term advantages, they will be equally vulnerable when political tides inevitably change. For decades, the defense sector was dominated by slow-moving, heavily regulated giants like Raytheon and Lockheed Martin for a reason. Functioning as an industrial arm of the Pentagon provided them the political insulation needed to avoid partisan battles, allowing them to concentrate on technology without restarting their strategy each time a new administration took office.

Today's startup rivals may be more agile than their predecessors, but they are far less equipped for the long haul.